holbrook huber 1979 jojs.pdf

MORRIS B. HOLBROOKJOEL HUBER

mE SPATIAL REPRESENTATION OF RESPONSESTOWARD JAZZ: APPLICATIONS OF CONSUMER

ESTHETICS TO MAPPING THE MARKET FOR MUSIC

Introduction: Consumer Esthetics

Whether the jazz artist's objective is to achieve mass appeal or to exposean uncompromised art form to the most receptive audience possible,success depends upon the design of an effective marketing strategy. Thedevelopment of such a strategy may be facilitated by the techniques ofconsumer esthetics. Some of these techniques, for instance, can be usedto generate spatial representations of the jazz consumer's perceptionsand preferences. In this article, such a model is described, categorized,extended, and illustrated by several examples based on empirical inves-tigations of the market for jazz.

Consumer esthetics, an emerging area of study within the broaderscope of marketing research, focuses on the buyer's cognitive, affec-tive, and behavioral responses to media, entertainment, and the arts.1The arousal of interest in such phenomena stems from the growingrealization that, like toothpaste or soda pop, artistic creations andperformances are "products" whose commercial success in the mar-ketplace depends directly on the way they are perceived and valued byconsumers.2 An analysis of consumer responses must therefore underlieany attempt to market esthetic products more successfully.

This viewpoint immediately raises the spectre-abhorrent to truedevotees of any art form-of "going commercial" or "selling out."The inevitable anti-commercial reaction depends, in part, on the fairlyplausible assumption that something complex and sophisticated cannot,by definition, be readily accessible to a mass audience-or else wewould change our definitions of complexity and sophistication-so that,

Morris B. Holbrook is assistant professor of marketing at Columbia University. Joel Huber isassociate professor of marketing at Duke University .Both authors have published numerousarticles on consumer research.

JOURNAL OF JAZZ STUDIES4

to achieve mass appeal, an artistic offering must perforce be watereddown and rendered simplistic.3 But this reaction assumes incorrectlythat the purpose of marketing is to confer mass acceptance on everyproduct it touches. In fact, the well-accepted principle of market seg-mentation suggests the converse: namely, that marketing strategyshould be designed to direct a given offering at the necessarily restrictedsegment of buyers whose needs and wants are best satisfied by theparticular configuration of appeals embodied by that offering.4 Viewedecologically, such a marketing orientation seeks the environmental"niche" that best reconciles the producer's strengths and weaknesses(in this case, the artist's talents, convictions, techniques, etc.) withexternal opportunities and threats (in this case, the limitations imposedby diverse patterns of audience tastes and preferences) .5 This means thatthe artist who wishes to create a deliberately esoteric and inaccessibleproduct can at least aim that offering at the audience most apt toappreciate it. Meanwhile, the more commercially oriented artist canrealize that no artwork can possibly appeal to everyone so that successon a large scale depends upon finding an especially large and receptivemarket segment and targeting one's marketing efforts accordingly.

Such considerations are particularly salient-tlnd sometimes evenpoignant--"in jazz, where the discrepancy between mass acceptance andrelative oblivion is probably as great as in any artistic field of endeavor .By focusing on the determinants of success in the jazz marketplace, thefield of consumer esthetics offers the performer, the producer, therecord manufacturer, the concert promoter, or the program directorinsights into the nature and structure of the consumer responses thatshape the fate of any jazz offering. Specifically, the techniques ofconsumer esthetics are well-adapted to the problem of analyzing andrepresenting the patterns of perceptions and preferences that influenceconsumer acceptance. The purpose of the present paper is to offer briefdescriptions of some of these techniques and to illustrate their applica-tion to studying the market for jazz. In particular , we shall focus on thoseprocedures designed to provide spatial representations of the relation-ships between important consumer variables. Such techniques are espe-cially interesting and useful because of the vividness and ea.sy inter-pretability with which they portray consumer perceptions and prefer-

ences in jazz.

RESPONSES TOW ARD JAZZ ;;; 5

A Taxonomy of Techniques for Spatial Representation in

Consumer Esthetics

In general, spatial representations of esthetic responses depict the psy-chological closeness between artistic objects-or between consumersand artists-by their distance from each other in a space of two or moredimensions. InterprQtation of the meaning of these dimensions may helpelucidate the underlying bases for esthetic response.

More specifically, techniques for forming esthetic spaces may beaistinguished according to ( I) their purpose and (2) their methodology.A cross-classification of these two distinctions produces the typologyshown in Example 1.

Ex. 1. A Taxonomy of Techniques for Spatial Representation in

Consumer Esthetics

Purpose

Perceptual Space(Perceptions Only)

Joint Space

(Both Perceptionsand Preferences)

Methodology

Study 1

Study 2

Study 3

Study 4

Compositional

Decompositional

As indicated in Example I, the purpose of a given technique may be( I) to form a perceptual space that indicates how some set of esthetic

objects is perceived by a sample of consumers or (2) to produce a jointspace that represents both perceptions and preferences in a way thatexplains feelings of liking or disliking for the objects. Both purposes areof clear importance to understanding the structure of consumer re-

sponses. Yet they deal with different facets of the problem.6 A purely

JOURNAL OF JAZZ S-tuDIES6

perceptual space simply represents cognitive responses toward artisticobjects, without attempting to indicate which objects are preferred. Ajoint space, on the other hand, usually models preferences by includingideal points that indicate the desired amounts of a given characteristicthat an object should possess in order to be regarded favorably. Relativepreferences among objects are then represented by their comparativedistances from such an ideal point.7

Example I also shows that the method of gathering and analyzing datamay be either (I) compositional or (2) decompositional.8 In brief,compositional approaches collect ratings of each esthetic object on anumber of attribute scales--e.g. , fast/slow, light/heavy, sweet/sour ,loud/soft, etc.9 These attribute ratings are then analyzed by some mul-tivariate statistical procedure such as principal components analysis(PCA) or multiple discriminant analysis (MDA) to create a space thatdepicts the differences between object perceptions by their locations ondimensions which condense and summarize the original attribute data.1o

By contrast, decompositional approaches begin by collecting judgmentsconcerning the perceived similarity between such possible pair of ob-jects. These pairwise similarity judgments are then analyzed by a tech-nique called multidimensional scaling (MDS) to infer the dimensionsthat underlie the original similarity ratings.11 Alternatively, mul-tidimensional scaling may be applied directly to preference ratings orrankings via a procedure known as unfolding analysis.12

The purpose of the present paper is not to provide detailed technicaldescriptions of these various multivariate statistical and scaling proce-dures.13 Our objective is to illustrate each approach as it applies tostudying the market for jazz.

The Spatial Representation of Responses to Jazz: SomeIllustrations

Study I: A Compositional Perceptual Space

In a carefully-controlled laboratory setting, 32 subjects (mostly graduatestudents and their compeers) listened to 15 unidentified tapes of jazzsaxophonists.14 These recordings are described in Example 2. Eachsoloist was rated on 18 semantic differential scales of the followingform:

oldnew

RESPONSES TOW ARD JAZZ

Ex. 2. Descriptions of J 5 Jazz Saxophone Recordings

Used as Stimulus Objectives in Studies J and 2

West Coast, Tenor

Lester Young

Stan Getz

Zoot Sims

Al Cohn

"Ad Lib Blues," D-flat, 222 beats/min. ,Verve VE-2-2502 1952

"Down Beat," B-flat, 238 beats/min. ,Verve MG V-8321 1957

"Zoot Swings the Blues," G, 262 beats/min. ,Prestige P-24061 1951

"John's Bunch," B-flat, 203 beats/min. ,Famous Door HL-107 1975.

West Coast, Alto

Paul Desmond

Lee Konitz

Art Pepper

E-flat, 210 beats/min. ,1963

B-flat, 108 beats/min. ,1956

B-flat, 200 beats/min.,1957

"St. Louis Blues,"

Columbia C2S-826

"Cork oN' Bib,"

Atlantic 1258

"Marty's Blues,"Archives of Jazz AJ510

East Coast, Alto

Charlie Parker

Sonny Stitt

Sonny Criss

"Jam Blues,"

Verve VE-2-2508

"Au Privave,"

Verve MG VS-6108

"California Screamin',Prestige 7628

"The Stanley Stomper,"

Prestige P-24065

F, 218 beats/min. ,1952

F, 259 beats/min.,1959


E-flat, 145 beats/min.,1956

Phil Woods

East Coast, Tenor

Sonny Rollins

Dexter Gordon

John Coltrane

Hank Mobley





"Bluesnote,"Blue Note BN-LA4O1-H2

"Wee-Dot,"Steeplechase SCS-lO25

"Some Other Blues,"

Atlantic 1354

"Walkin',"Columbia C2S-82O

8 JOURNAL OF JAZZ SRJDIES

The bipolar adjectives used to construct these scales were as follows:new/old, feminine/masculine, traditional/contemporary, slow/fast,active/passive, busy/lazy, composed/improvised, emotional/intellectual, out-of-tune/in-tune, changeable/stable, complex/simple,predictable/unpredictable, random/structured , repetiti ve/ shifting ,light/heavy, dissonant/consonant, warm/cool, and poorly-recorded/well-recorded.15 The saxophone improvisations were each based uponthe major 12-bar blues form and consisted of three choruses from themiddle of a solo so that easily recognized tunes were masked. Otheraccompaniment, in all cases, was provided by just a rhythm section:bass, drums, and piano or guitar. Thus, musical form and instrumenta-tion were held constant while other stylistic considerations were per-mitted to vary .For instance, as shown in Example 2, an effort was madeto include both tenors and altos, both East and West Coast orientations,and a wide spectrum of keys, tempos, and recording dates.

As a first step, the data were submitted to a principal componentsanalysis, which suggested that they could be well summarized by thefollowing six indices: activity, harmoniousness, newness, orderliness,softness, and masculinity .16 Scores on these indices were then enteredinto a multiple discriminant analysis to generate the two-dimensionalMDA space shown in Example 3.

The horizontal dimension in Example 3 clearly represents tempo-ranging from slow (Konitz) to fast (Stitt)-while the vertical dimensionreflects the distinction between East and West Coast orientations, withRollins, Gordon, and Coltrane at one end of the continuum and Young,Desmond, and Sims at the other. Admittedly, there is some overlap ofstylistic groups at the center of the vertical axis. Thus, for example, thespace might be more pleasing intuitively if the vertical positions ofKonitz and Parker were interchanged. But such apparent anomalies onlyserve to underscore the important point that the space is in no wayintended as a substitute for musicological analysis. Rather it representsthe perceptual judgments of a particular sample of jazz consumers,whose ratings may reflect attention to details that a musicologist wouldnot necessarily regard as important aspects of style. It is, however ,consumers and not musicologists who spend money for record albumsso that, in a marketing-oriented approach, it is consumer perceptionsthat matter most.

RESPONSES TOW ARD JAZZ 9

Ex. 3. MDA Space for J 5 Saxophonists in Study J

IIIIIIIIIIIIII

Youn~Desmond !

SimsII

Ge~zI

Pepper .Parker I StlttI .

Cohn : Crlss

I~

Rollins

Gordon

StUdy 2: A Decompositional Perceptual Space

The same group of subjects used in Study 1 also provided six-pointsimilarity ratings for each pair of saxophone recordings. Specifically,they listened to one recording, then the next, and rated the pair forsimilarity-continuing until all 105 pairs had been judged. Such exten-sive listening-based comparisons are, we believe, unique in experi-mental esthetics because of the enormous amount of time required tocomplete all 105 pairwise comparisons.17 The advantage of such alaborious procedure, however, is that it forces the subject to rate un-

~ ,--

+-\A Konitz : Mobley

IIWoods I

,,,,II

ColtraneI

JOURNAL OF JAZZ STUDIES10

Ex. 4. MDS Space for 15 Saxophonists in Study 2

II,,IIIIIII

Pa,'rkerI

Cohn

Desmond

Pepper Young

Coltran~etz

IIIIIII

S .I lmsIIII

~ MobleyI .

: CrlssStittIIIIIII

: GordonIII1.11IIII.

!

Rollins

Woods

Konitz

i

identified selections on the basis of direct experience with the musicrather than on the basis of any extra-musical associations.

When these pairwise similarity judgments were submitted to mul-tidimensional scaling, the resulting two-dimensional MDS space(Example 4) resembled that found in Study 1 in most respects.18 Forexample, the correlation between the spaces across the 105 inter-pointdistances was r = .71.19 This moderately high value for r reflects the fact

that the horizontal axis can still be interpreted as representing tempowhile, in general, the vertical dimension still represents the distinctionbetween East and West Coast styles. The departure of r from itsmaximum value of 1.0, however, results from the fact that, onmusicological grounds, Konitz and Parker are again misplaced, muchmore seriously here than in Study 1.

RESPONSES TOW ARD JAZZ

It must be reiterated that, because the data represent consumer per-ceptions, the "misplacement" of musicians can be inferred only byimposing some external musicological standard. Thus, the position ofKonitz, though unexpected, is not necessarily erroneous perceptually.Indeed, he was the only saxophonist accompanied by guitar (BillyBauer) instead of piano. If this aspect of the Konitz recording wasimportant enough to exert a strong effect on his similarity ratings, itmight explain his position at some distance from the saxophonists withwhom he is normally associated. The absence of such a pronounceddiscrepancy in Study 1 could therefore have resulted from the fact thatno piano/guitar attribute was included in the ratings collected in thatstudy. Thus, though the space in Study 2 (Example 4) may seem lessplausible than that found in Study 1 (Example 3), it may be moreaccurate in a psychological sense since subjects were free to make theirown similarity judgments independent of any narrower set of considera-tions artificially imposed by the experimenters.

Study 3: A Compositional Joint Space

In a third study, a similar but independently drawn sample of 16 subjectsrated 15 saxophonists on a set of 93 semantic differential scales compa-rable to those of Study I, but much more extensive in scope. The keydifference in Study 3, however, was that, in place of Hank Mobley,subjects rated a hypothetical ideal saxophonist on the same set of 93attributes.2o This made it possible for principal components analysis,followed by a subsequent multiple discriminant analysis of the factorscores, to create a two-dimensional space representing both the per-ceived positions of saxophonists and the location of an ideal point foreach subject.

The resulting MDA space appears in Example 5. Again, the horizon-tal dimension offers a clear representation of tempo, but here the verticaldimension emphasizes the distinction between altos and tenors (withsome tendency for East Coast players to be placed lower and West Coastplayers higher within their instrument groups). Moreover, the widedispersion of the ideal points (I1 to 116) offers evidence of considerableheterogeneity of preferences among the subjects. Indeed, the assump-tion that different consumers have different needs and wants is funda-mental to the concept of market segmentation and appears to be sup-ported by the wide scatter of ideal points found in Example 5.21

JOURNAL OF JAZZ S'ruDIES12

Ex. 5. MDA Space for 14 Saxophonists and Ideal Points in Study 3

I

: 114

DesmpndII

: 11I I

5pa~ker

PepperI16! Stitt

111:II

: GetzKonitzI 1 12 I Cri~~ O ~ w ,

I Woods Young 'Coltrane15 Co~n

16 sJms

III

14 !

113:II

I !912

RollinsGordon

3 110

1

Study 4: A Decompositional Joint Space

An example of the fourth type of spatial representation used a verydifferent, and much less jazz-oriented, set of esthetic objects consistingof 12 female singers: Cher Bono, Ella Fitzgerald, Roberta Flack, CaroleKing, Barbra Streisand, Diana Ross, Bette Midler, Aretha Franklin,Liza Minnelli, Carly Simon, Pearl Bailey, and Dinah Shore.22 These 12singers, plus a l3th ideal singer, were rated for pairwise similarity by acomparable but independently drawn sample of 22 subjects. The result-ing MDS space appears in Example 6.23

Here, the horizontal axis appears to represent ethnic membership(white versus black) while the vertical dimension reflects degree of


Ex. 6. MDS Space for 12 Female Singers and Group Ideal Point inStudy 4

IIIIIIIIIII

King :

IIIII

,

Simon

GroupIdeal

: Flack

Midler : Ross

Streisand : Franklin, ~ I

Minnelli :

IIIIIIIIIIIIIIII

: FitzgeraldII

Shore IIII ..I .!

Bailey

contemporaneity (old versus new) .Moreover, this particular approachcreates a group ideal point close to Roberta Flack in the contemporaryblack region of the space. If desired, respondents could be clustered intoaffectiveiy homogeneous groups before submitting the data to mul-tidimensional scaling so that the resulting group-based MDS spacescould be viewed as market segments with shared preferences and acommon ideal point.24 Alternatively, unfolding analysis could be usedto identify individual ideal points comparable to those shown in Exam-pie 5.25 The authors believe, however, that a more fruitful approach tosegmentation analysis is provided by the extension described in the nextsection.

14 JOURNAL OF JAZZ sTuDIES

An Extension of the Spatial Representations of EstheticResponses: The Method of Multidimensionally ScaledCorrelations

The four types of spatial representation described ab9ve, though illus-trated in a laboratory setting, could in principle be applied to large-scaleconsumer samples. A partial limitation of the foregoing examples,however, is that they offer few insights into how the jazz market can besegmented or what segments tend to prefer which jazz artists. Suchanalyses could be handled by a number of available procedures,26 but anapproach directly related to the aforementioned spatial analyses is themethod of multidimensionally scaled correlations (MSC).27

Multidimensionally Scaled Correlations (MSC)

Typically, MSC begins with data representing each respondent's prefer-ences toward (or involvement with) a number of esthetic objects. Thesemeasures are correlated (between objects and across respondents) toprovide a set of pairwise measures of association that are submitted tomultidimensional scaling. In the resulting MDS space, therefore, inter-point distances represent the degree of preference association between

objects.Next, the original preference scores are correlated with general cus-

tomer characteristics such as demographic variables (age, sex, ethnicgroup), socioeconomic status (income, education, occupation), orpsychographic variables (personality, life style). The correlations of anycustomer characteristic with the set of object preferences are thenrepresented in the space by a vector running through the origin.28 Theinterpretation of such a characteristic vector is straightforward andextremely interesting: the farther an object is located in the positivedirection of the vector, the more positively preference for that object isassociated with possessing the characteristic represented by the vec;tor .In this way, the set of vectors provides a vivid representation of theoverall pattern of relationships between esthetic preferences and thepossession of various demographic, socioeconomic, and psychographiccharacteristics .

Study 5: The WRVR Survey

This approach is illustrated by a study conducted for WRVR, the major

SPONSES TOW ARD JAZZ 15

jazz radio station in New York City .29 Historically, the station hadn shifting its programming format from a focus on mainstream jazz ,yed by disk jockeys who provided detailed information about each)rding, to an increased emphasis on crossover music announced byker, more personality-oriented DJs. These programming revisionse intended to win a larger audience for the station, but an appraisal ofsuccess of the format changes depended upon an assessment of thecture of the market for jazz. This structure was by no means self-lent to WRVR's management.'s a basis for undertaking such a market analysis, mail-survey ques-naires were distributed to 1,000 and returned by 327 respondents.~e questionnaires collected 9-point preference ratings of 9 radioons, 14 local disk jockeys, and 59 we11-known jazz musicians::ted, with the help of WR VR ' s management, to represent a diversity

lZZ styles and periods.30 In addition, the questionnaires obtainedsures for a number of general customer characteristics such asD graphics, socioeconomic status, record-buying activity, and life-31

ultidimensional scaling of the inter-musician and inter-DJ corre-tiS generated the MDS preference space shown in Example 7.32 Inspace, the positions of the musicians (Small Letters) and DJs>IT ALS) suggest an almost unequivocal interpretation of the twoVertically , there is a clear distinction between familiar , widely-

sed performers at the top (Quincy Jones, Ray Charles, Miles Davis,; Armstrong) and more unfamiliar , esoteric artists at the bottomRa, Marion Brown, Gary Burton, Hampton Hawes). Horizon-the preference space separates the more commercially successful:ians at the left (George Benson, Herbie Hancock, Freddie Hub-Donald Byrd, Chuck Magione, Grover Washington, Hubert

, Chick Corea, Bob James) from the more purist musicians on theNho are more faithful to their jazz heritage (Dizzy Gillespie, Oscaron, Charlie Parker, Earl Hines, Lester Young, Paul Desmond,rd Brown, Barney Kessel, Zoot Sims, Gerry Mulligan, Joe Pass,~pper). Subjectively, the pattern of positions in the preferencehangs together intuitively with a pleasing degree of consistencyrarely found in consumer research-perhaps because judgments,umer esthetics are made with more confidence, involvement, andise than are evaluations of products like toothpaste or cigarettes.positions of the disk jockeys in the preference space are equallyable. Max Cole and Ed Beach had been noted for playing slightly

AZz:STUDIESJOURNAL OF16

Ex 7. MDS Preference Space for 59 Jazz Musicians (Small Letters)

and 8 WRVR Disk Jockeys (CAPITALS) in Study 5

FAMILIAR

~harles ,

: Armstrong

,.., Davis Ellington: Goodman Basie

ALTMAN :Brubeck

Q Jones

Ferguson

!nson bllleSpie

'eterson

Parker

Hines p

Monk U

Young

Desmond C Brown R

I---T~tum Ke~selBEA(;H Slms S

.COLE Mulligan T

:kson Pass

Farmer "t PepperKonl z

B EvansHall

Woods

Coltrane

IIII

DAVI~ MJI

cD Byrd Hancock

() ALEXANDER Hubbard

M Manglone

Washington

M A H Laws Coreayers

E James

R R [aws Jarreau HARRIS

C HERSCHEL

I Klerrmer DiMeola

Pnntv

Mingus

.lac

AIIIII ColemanII

Cl arke 'J"~'

anborn

McLaughlinAirto Sanders

Jarrett MAXANNE

M Brown

Sun Ra Bu~onIIIIII

ESOTERI (

Hawe

obscure musicians committed to the jazz tradition. By contrast, G. KeithAlexander, Doug Harris, and Herschel all featured rather commercial,pop-oriented performers and would therefore presumably have beenenjoyed most by those whose tastes ran in the direction of mass-appealcrossover music. Maxanne' s chief characteristic at the time of the studywas that she was new on the station in the after-midnight time slot.Accordingly, she was located at the unfamiliar end of the verticaldimension close to the musicians preferred by those whose tastes weresomewhat esoteric. Roberta Altman was high on familiarity, but neutralon commercialism--perhaps because her role consisted mostly of read-ing the news on the morning show with Les Davis, to whom she wasadjacent in the space. Most interesting was the completely neutral

17)NSES TOW ARD JAZZ

8. Vectors Representing General Customer Characteristics in

the Preference Space in Study 5

0

'b

.(.'b

~'?:;;-

"v ~~c, vq,

~ q,<:'

fI."~fl.q,'/ .(q,0

:b"J

~c;.0(;

, ~C'o~ ~~.1."Jt ~}.f: J't ".I\~ t 0;, -r~

,. 0/) ~J'~OJ- "0

1tt8t.S ~. Q Jones

61QC/r ""rerguson

7IQle ,. D Byrd Hancock -

1 SCO S ALEXANDER Hubb.r,-Mang1one Washington

A~e?~WS Corea

JamesI'.i:; .R~i~~~;u---,

HERSCHEL ,~Klenmer DiMeola ,

.1\\10\1S-- Ponty\ .Clarke\\1S"\ C Sanborn ...'

McLaughlinAirto Sanders

/ Jarrett MAXANNE., -" M Brown

c , -.Sun Ra

/:: I '

Charles :/ ,

: Armstrong~ Oavis EII;ngton: Goodma7 Baste O

AlT""N :Brubeck , 'fe~ .~s: G;llesp;e ./ e~'e! Coltrane Peterson /' 'f

.' park~r ge./ Hines '.1" jaZZ

I'onk ::::---;<?.

I'J~ Young i "come, Oes""nd C Brown a rri ed ,

TB ~~ Kes$el OW"~ ' ~Ol~ S,ms. records

Jac~ Mul1,gan

Farmer Pass edu~ ' Kon;tz pep~~ cationB Evans ~Q 1P. Ste ,

Colerr~n Hall, e Y'eoWoodsH.wes "' " \ .,

, ~- "

// Tyner

,~I;)

(~~.;o t~., ~ 4'1ItC'1It

~,~~ ~,~(, ,~~

~o ~"o,~

Buriton

\

\~.-k{0

,..>.9'

~~

-Q;rt 1:

Q1",-QI'

ition of Les Davis himself. This may be explained by the fact that hele, among the old-guard WRVR disk jockeys, was attempting topt in the direction toward which the station' s programming was{ing. One might speculate that, over time, his preference position, shifting from a location near to Max Cole and Ed Beach toward at closer to a. Keith Alexander and that, at the moment of the study,was therefore caught on neutral ground in the preference space.'he next step was to place vectors representing general customerracteristics into the preference space.33 The set of such vectors ofItest interest to management is depicted in Example 8. Here it can be1 that enjoyment of WRVR was positively associated with prefer-es for more falniliar-and-commercial musicians such as Maynardguson, George Benson, Herbie Hancock, Donald Byrd, and Chuck

18 JOURNAL OF JAZZ-'STUDIES

Mangione as well as with a greater liking for the newer WRVR diskjockeys like G. Keith Alexander, Herschel, and Doug Harris. Thispattern of tastes also characterized enjoyment of WBLS-the soul-music station that management viewed as WRVR's key competitor,especially since the two were located in adjacent spots at one extremeend of the FM dial. By contrast, Max Cole and Ed Beach (old-guardWRVR DJs) tended to be admired by fans ofWNCN. The latter was aclassical music station, also located close to WRVR on the dial. Perhapscoincidentally, Max Cole went to work for WNCN after leavingWRVR.

In terms of demogra,phic and socioe~onomic characteristics, Example8 also shows that preference for the more purist artists was associatedwith being older, higher in income, married, well educated, male, andwhite. By contrast, liking the more commercial musicians was associ-ated with the opposite general customer characteristics. Taken together ,these relationships suggest a clear tendency for WRVR's new pro-gramming format to appeal more to a socioeconomically "mid-scale"audience whereas its old format had attracted a relatively "overprivi-leged" listenership. By implication, this "up-scale" segment would bealienated by the station's shift toward more crossover programming.

Finally, Example 8 portrays a subset of the 46 life-style and purchas-ing vectors selected to suggest the nature of the related musical activitiesengaged in by those preferring musicians in different regions of thespace. Those preferring familiar commercial musicians, for example,tend to buy their records at discount stores, to favor top-40 music, and tovisit discotheques. Those who like more esoteric, but commercial,artists tend to seek continuous music by carrying a portable radio aroundwith them. In sharp contrast, those favoring familiar purist jazz playersappear to be better informed-tending to read reviews and music maga-zines, to develop firm musical preferences, and to prefer records toradio. Those who enjoy the purist, but more esoteric, musicians aremore likely to go out to hear live jazz and to buy their records at bigrecord shops like Sam Goody or King Karol. But probably the mostloyal market for jazz recordings appears more among those whosepreferences lie toward the extreme end of the purist dimension: they aremore likely to own records in general, jazz records in particular, andexpensive stereo equipment.

In sum, then, the preference space and characteristic vectors indicatethat WRVR 's programming change had achieved its intended objective.The station had become relatively more popular with those favoring the


mass-appeal artists in the familiar commercial region of the space, at theexpense of alienating those preferring musicians with more limitedappeal. Clearly, this latter group constituted a loyal, relatively affluentsegment with more narrowly defined jazz tastes. Certainly, it was asmall segment in terms of absolute size. Yet, from an advertiser's pointofview, a manufacturer of records or hi-fi components might well havewanted to target his promotional messages at precisely this audiencebecause of its comparative willingness and ability to expend financialresources on music. With the shift in WRVR programming toward theupper left -hand comer of the preference space, that option was no longeravailable to advertisers. This illustrates the point that undifferentiatedaudience ratings, such as those provided by Arbitron, may be quitemisleading to an advertiser trying to reach a particular market segment.In the present case, by appealing more broadly to the mass audience thatlikes familiar commercial music and thereby improving its Arbitronratings, WRVR may have rendered itself less useful as a specializedvehicle for selectively reaching the hard-core jazz fans who spend themost money for records and equipment.

Discussion

The present paper has not attempted to resolve the thorny question ofwhether the jazz artist should or should not compromise the art form inorder to seek a greater degree of popularity. Rather the purpose has beento illustrate research techniques that permit a particular jazz offering,however uncompromising, to reach the most favorably disposed audi-ence given the limits imposed by the nature of the marketplace. Thoselimits are best understood in terms of the patterns of perceptions andpreferences that characterize consumer responses to jazz. Such analysesof the structure of consumer responses can often be greatly facilitated bythe techniques of spatial representation exemplified in the five studiessummarized here. These spatial techniques should prove useful in theempirical work needed as a basis for the rapidly emerging study ofconsumer esthetics.

20 JOURNAL OF J~Z STUDIES

NOTES

i. Morris B. HoIbrook and Douglas V. Holloway, "Patterns of Relationships Among EstheticPreferences and General CuStomer Characteristics: An Application of Multidimensionally ScaledCorrelations to Mapping the Market for Music," working paper, Columbia University (1978),available from authors.

2. Morris B. Holbrook and Joel Huber, "Separating Perceptual Dimensions From AffectiveOvertones: An Application to Consumer Esthetics,'. Journal of Consumer Research 5 (March1979): 272-83.

3. For an impassioned journalistic statement of this viewpoint, see Gene Lees, "The Dotage ofAmerican Radio," High Fidelity (January 1978): 12-29. A careful sociological formulation of asimilar argument is provided by William N. McPhee, "Natural Exposure and the Theory ofPopularity," inFormal TheotiesofMassbehavior (New York: The Free Press, 1963), pp. 104-68,and tested in a marketing context by Abe Shuchman;" Are There Laws of Consumer Behavior? ,"Journal of Advertising Research 8, no. I (March 1%8): 19-27.

4. Ronald E. Frank, William F. Massy, and Yoram Wind, Market Segmentation (EnglewoodCliffs, N.J.: Prentice-Hall, 1972); Morris B. Holbrook and John A. Howard, "Frequently Pur-chased Nondurable Goods and Services," in Robert Ferber, ed., Selected Aspects of ConsumerBehavior (Washington: National Science Foundation, 1977).

5. Wroe Alderson, Marketing Behavior and Executive Action: A Functionalist Approach toMarketing Theory (Homewood, Ill.: Richard D. Irwin, 1957).

6. Indeed, became one deals primarily with perceptions while the other focuses on feelings, thedimensions underlying the two kinds of space do not necessarily coincide; see Holbrook and Huber ,, , Separating Perceptual Dimensions, ' ,

7. Paul E. Green and Yoram Wind, Multiattribute Decisions in Marketing (Hinsdale, Ill.: TheDryden Press, 1973); Holbrook arid Howard, "Frequently Purchased Nondurable Goods andServices."

8. In addition to Green and Wind, Multiattribute Decisions in Marketing, and Holbrook andHuber, "Separating Perceptual Dimensions," see William L. Wilkie and Edgar A. Pessemier,.'Issues in Marketing'stJse ofMulti-Attribute Attitude Models," Journal ofMarketing Research10 (November 1973): 428-41.

9. For development of the theory underlying such scales, see Charles E. Osgood, George J. Suci,and Percy H. Tannenbaum, The Measurement of Meaning (Urbana, 111.: The University of Illinoispress, 1957).

10. These uses o£PCA and MbA are discussed, respectively, by Glen L. Urban, "Perceptor: AModel for ProductPositioiling," Management Science 21 (April 1975): 858-71, and by Richard M.Johnson, "Market Segmentation: A Strategic Tool," Journal ofMarketing Research 8 (February1971 ): 13-18; a detailed technical <:omparison of the two procedures is provided by Joel Huber andMorris B. Holbrook, "Using Attribute Ratings for Product Positioning: Some Distinctions Be-tween ComPOsitional Approaches," Journal ofMarketingResearch 16 (November 1979, in press).

II. MDS techniques are clearly described by Paul E. Green and Frank J. Carmone, Multidimen-sional Scalilig and Related Techniques in Marketing Analysis (Boston: Allyn and Bacon, 1970),and by Joseph B. KruskaI and Myron Wish, Multidimensional Scaling (Beverly Hills: Sage

Publications, 1978).


12. C. H. Coombs,A TheoryofData (New York: John Wiley, 1964); Paul E. Green and Frank J.Carmone, ' 'Multidimensional Scaling: An Introduction and Comparison of Nonmetric Unfolding

Techniques," Journal of Marketing Research 6 (August 1969): 330-41.

13. For such technical descriptions, see the references cited in Notes 1,2,7,8, 10, II, and 12.

14. Morris B. Holbrook and Joel Huber, "Musicological Distinctions Between Jazz Styles: CanAnybody Really Hear the Difference?" (in preparation).

15. These 18 adjectival pairs were selected to represent the dimensions shown to be important by

Holbrook and Huber, .'Separating Perceptual Dimensions."

16. These indices were composed by combining scores on the adjectival attributes in a mannersuggested by the PCA results: activity = fast -passive -lazy; harmoniousness = in-tune +

consonant + well-recorded -heavy; newness = contemporary -old; orderliness = stable +

simple + structured -unpredictable -shifting -improvised; softness = intellectual + cool; and

masculinity = masculine.

17. To avoid fatigue effects, the ordering of pairs was carefully balanced and rotated and,therefore, required that each subject listen individually.

18. Specifically, MDS was performed by the INDSCAL procedure derived by J. D. Carroll and J .J. Chang, " Analysis of Individual Differences in Multidimensional Scaling Via an N-Way

Generalization of 'Eckart-Young' Decomposition," Psychometrika 35 (1970): 283-319, anddescribed by S. Pruzansky, "SINDSCAL: A Computer Program for Individual Differences inMultidimensional Scaling," Bell Laboratories (1975).

19. The correlation coefficient, r. ranges between -1.0 and + 1.0 and describes the degree ofassociation between two variables.

20. Holbrook and Huber, "Separating Perceptual Dimensions." The rating of a hypothetical idealobject is a device widely used in consumer research on brand preferences. It asks the respondent toindicate a most-desired position on each attribute scale, thereby permitting a measure of the extentto which every product offering is perceived to depart from this ideal level.

21. The validity of the positions of these ideal points is indicated by the relatively good fit obtainedwhen the distance between a subject's ideal point and a saxophonist's position was used to predictpreference, as measured by a summative index of global evaluation: mean adjustedR = .47. Giventhe sample size (N = 16) , this fit was statistically significant at p < .01 .

22. Morris B. Holbrook and Rebecca S. Williams, ..A Test of the Correspondence BetweenPerceptual Spaces Based on Pairwise Similarity Judgments Collected With and Without theInclusion of Explicit Ideal Objects," Journal of Applied Psychology 63 (1978): 373-6.

23. Again, the INDSCAL procedure was used.

24. This approach to segmentation is described by Lester A. Neidell, "The Use of NonmetricMultidimensional Scaling in Marketing Analysis," Journal of Marketing 33 (October 1969):37-43. It is important to note that, whereas the individual ideal points in Example 5 r~present most-desired positions for each respondent considered separately, the' , group ideal' ' in Example 6 is an

aggregate indication of the consensus across subjects.

25. In addition to Paul E. Green and Frank J. Cannone, "Multidimensional Scaling," see J.Douglas Carroll, "Individual Differences and Multidimensional Scaling," in Roger N. Shepard,

JOURNAL OF JAZZ STUDIES

"d AppliIg: Th.~ultidimen,

ninar Press

A. Kimball Romney, and Sara Beth Nerlove, elcations in the Behavioral Sciences (New York

Holbrook and Howard~6. Frank, Massy, and Wind, Marke:based Nondurable Goods and Servic

'gmentatj

!8T"h""'lly'h'...";o"Of'",h"'ct°"'""'mi"'dbYre8re,,;"8'h"h="'ri""'0""";0"' w;.h'h,obj~'pref're",',""'h"p,.;,I,00'",","'of'h'obJ"""'o.d;"gw'h'p,oc,dore rugg,"od " ", b,,;, fu, PREFMAP by C,""llfud"'"o,] D"re"",,, ,"d Mo"

;dim,"';0",IS,,];"g

Relationship!)Ibroc and Hc attem

enjoyableenjoyable'he 9-point pre

e given by Holbrook and Holloway, "Patterns

provided by William D. Wells, "Psychographicsearch 12 (May 1975): 196-213.

these measure

:neral discussio

r of Marketing

31. Specific ex

Relationships."Critical Review

amples of

A more ge

," Journal R,

rocedure used here

'How To Use KYS'Bell Laboratories I

described by Joseph B. Kruskal, Forrest W. Young, and

A Very Flexible Program To Do Multidimensional Scaling

973).

32. Thf

Judith E

and Unl

YST

eery

ling,

m procedure mentioned in Note 28. Onl:

.70 were regarded as sufficiently reliabt

ial interest to WRVR's management-i

.This was accomplished using the multiple regressi)se vectors with multiple correlations greater than R ,

be retained. Of these, only a selected set---0f spe,

:scribed here.

holbrook huber 1979 jojs.pdf

Documents